Streamline News

Integrating Repository Function with Work Practice

Aquabrowser

Posted by Nick on September 17, 2008

I mentioned in the meeting yesterday that Streamline may want to look at Aquabrowser which is a novel resource discovery platform that the library has been looking at recently.

The screenshot below is from a library site running the software and shows results for “fox”; on the right hand side results are returned in a traditional manner; the “discover” pane on the left hand side is the clever bit which positions your search string at the centre of a colour coded web of association; spelling variation; translation; thesaurus terms; discovery trail.  A user can click on an association which is pulled into the center of the web; the search results on the right are modified accordingly so if I click on “desert” my results comprise “The desert fox” and “The Black Rock Desert” by William. L. Fox etc

To have a go visit:

http://aqua.queenslibrary.org/

Advertisements

3 Responses to “Aquabrowser”

  1. johnrg said

    Hi Nick,
    An interesting tool. I have been contemplating using the meta data returned along with each hit on a search term and using these to build a ‘Tag Cloud’ to show related meta data for hits – it would at least show both what meta data is attached to each hit and the relative frequency of these among the search hits i.e. if you found these with the original search term you might want to expand the search to include these terms as well.

    I’m not exactly clear where Aquabrowser gets the terms it uses to include in its discover pane but it is an interesting approach.

  2. Nick said

    Hi John

    We (Dawn, Mark, Mike and me) had an interesting meeting yesterday at which Mark introduced 4 ideas for augmented search – I shan’t try to describe them all here – I wouldn’t do them justice and I’m sure Mark will disseminate them in due course – but the simplest was “pre-search association” whereby a search string is put through a thesaurus to generate secondary search terms – some of his other ideas were much more sophisticated – then I guess we’d need some sort of GUI to visualise associated results (a tag cloud?).

    It also occurs to me that users are able to tag resources themselves in intraLibrary – could these tags be used in a way that was diffeerentiated from ‘formal’ metadata to produce, I dunno, some sort of user-generated associations?

  3. johnrg said

    Hi Nick,
    Picking up on Mark’s idea if we look at pre-search association then if we used the same approach applied in generating meta data to standardise synonymous words etc. presumeably we would be able to identify words close to the search term? A ‘Tag Cloud’ very largely offers a visual representation of the frequency of words within a document or collection i.e. it reads the source and counts words, using their frequencies to present the words in different sized fonts, the bigger the font the greater the frequency.

    As ever the value of this approach is dependent on the accuracy / correctness of the source used; I think that it would be very useful to see if we can create a prototype tool that generates a tag cloud and see how helpful it is to users. One criterion could be to do a search for a library book by using the library catalogue, going to the relevant shelf, finding the book and then looking around nearby books to see if there are any other books that are equally useful but which didn’t appear in the search resuls. Then see if we can do the same thing but build a ‘tag cloud’ from the search results and see how that maps onto the physical book search….

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

 
%d bloggers like this: